Optimization of a photochemical gold nanoparticle synthesis

Transcription

1 Optimization of a photochemical gold nanoparticle synthesis B. Büttner*, J. Ott*, C. Bartzsch*, S. Kunath**, M. Schimmelpfennig**, K.-H. Feller* *Ernst-Abbe-Hochschule Jena Institut für Mikrosystem- und Präzisionsfertigungstechnik Jena **Dynardo GmbH - Weimar 1

2 outline workgroup instrumental analysis motivation photochemical synthesis of gold nanoparticle pre optimization DoE I optimization DoE II comparison experiment and simulation; outlook 2

3 workgroup optimization tasks simulation experimental biological chemical 3

4 outline workgroup instrumental analysis motivation photochemical synthesis of gold nanoparticle pre optimization DoE I optimization DoE II comparison experiment and simulation; outlook 4

5 motivation nanoparticle are ingredients in products of daily use color and intensity must be constant 5

6 What are Nanoparticles? particle size: nm mostly artificial origin first usage stained glass Khlebtsov, N. G. and L. A. Dykman (2010). "Optical properties and biomedical applications of plasmonic nanoparticles." Journal of Quantitative Spectroscopy and Radiative Transfer 111(1):

7 nanoparticle material metal nanoparticle noble metals metaloxids semiconductor Au Ag ZnO Fe 2 O 3 CdTe CdSe 7

8 Surface Plasmon Resonance Spectra Nanoparticles have specific colors Surface Plasmon resonance is size and charge dependent different particles have a huge shift in color amount of particle change color intensity samples of different Gold- Nanoparticle sizes with specific color changes 8

9 outline workgroup instrumental analysis motivation photochemical synthesis of gold nanoparticle pre optimization DoE I optimization DoE II comparison experiment and simulation; outlook 9

10 photochemical synthesis photoinitiator cleavages radicals under light- irradiation light intensity controls reaction needs high energy (UV-region of spectra) reaction is easily stoppable microfluidic set up can increase yield radicals reduce Au ions colloidal gold is produced Martin et. al,

11 raytracing simulation of set up setup with optimized optical irradiation cuvette UV LED 11

12 time resolved experiment NP- synthesis take around 5 min to start size of particles changes with reaction time and reactant concentration Time dependent uv-vis spectra of NP-synthesis with I

13 outline workgroup instrumental analysis motivation photochemical synthesis of gold nanoparticle pre optimization DoE I optimization DoE II comparison experiment and simulation; outlook 13

14 parameters Parameters AuCl4 Irgacure 2959 irradiation time 14

15 extinction [A.U.] particle analysis UV-VIS spectra peak height [A.U.] = particle amount wavelength [nm] = size full width half maximum (FWHM) [nm] = size distribution wavelength [nm] 15

16 method procedure perform first experiments to get an idea for the: input factors controllable non controllable: collect every possible measured values! output factors design a first design of experiments to: verify the factors and factor levels get a first idea of the system behavior detect regions in design space with failed designs second design of experiment with the aim: move design space to regions without failed designs to verify model of the system to increase dataset in the volume of interest 16

17 method Central-Composite-Design method of choice for discrete input parameters with 2-3 factor levels easy to extend robust against failed designs information concentrates to the boundary of n-dimensional hypercube linear and quadratic approximation functions useable no random experiments Central- Composite- DoE scheme optislang Documentation 17

18 preoptimized system 3 ary Central Composite Design with 15 experiments (13 succeeded) gives a hint about system behavior Visualization of dependency between NP-size and input parameters (CoP 60 %) 18

19 preoptimized system II temperature dependency's difficult to interpret extinction wavelength FWHM CoP- matrix for DoE I 19

20 extinction at 520 nm [A.U.] Michaelis constant temperature dependency test reactions are a result of better system understanding synthesis yield independent from temperature reaction kinetics increases with temperature contradictory to chemical reaction kinetics extinction Michaelis constant regression line temperature [ C] 20

21 outline workgroup instrumental analysis motivation photochemical synthesis of gold nanoparticle pre optimization DoE I optimization DoE II comparison experiment and simulation; outlook 21

22 Design of Experiment II Central-Composite-Design 33 experiments moved design space controllable Parameters non controllable AuCl4 Irgacure 2959 irradiation time temperature 0,15mM ratio 3 5min start 0,3mM ratio 5 20min max 0,45mM ratio 6 35min end 0,6mM mean input parameters with levels Standard derivation 22

23 meta modelling Should more than one of the temperature parameters be used? max-temperature SD-temperature mean-temperature end-temperature start-temperature Correlation matrix for temperature input parameter No, input-correlation between these parameters are to high 23

24 meta modelling Meta-Model-of-optimal Prognosis Which input parameter should be used? temperature parameter Wavelength [nm] particle size Full Width Half Maximum [nm] Size distribution Extinction [A.U.] Particle amount none 83,0 87,9 94,8 all 73,1 72,5 92,8 start 78,9 88,3 93,6 max 80,2 88,2 93,7 end 79,4 87,0 92,6 mean 80,2 88,2 93,7 standard derivation 79,6 81,0 94,5 COP- values in dependence of different temperature input parameter 24

25 meta modelling meta models with high quality can be generated behavior of system can be read out of the approximations 25

26 objectives size distribution -> min particle amount -> max irradiation time -> min optimization process different optimization algorithms possible huge differences in computational effort correlation matrix Particle swarm optimization evolutionary algorithm population size: Minimum generations: 5 5 Maximum generations: comparison parameter for optimization algorithms 26

27 evaluation of optimal synthesis evolutionary algorithm needs more data points slight differences in optimal synthesis Pareto-front PSO-algorithm Pareto-Front EA-algorithm Particle swarm optimization evolutionary algorithm irradiation time [min] Concentration Auric acid [mm] 0,58 0,6 ratio auric acid and photoinitiator 4,3 4,8 27

28 evaluation of optimal synthesis evaluation shows stable system acceptable deviations to prediction except of wavelength /size extinction 2,3 wavelength 521,6 FWHM 51,18 Particle swarm optimization prediction evaluation evolutionary algorithm prediction evaluation Extinction [A. U.] particle amount 2,2 2,3 2,8 2,53 ± 0,12 wavelength [nm] Size 522,3 521,1 ± 0, ,6 ± 0,26 Full Width Half Maximum [nm] Parameter bounds for EA-algorithm size distribution 52,3 53,1 ± 0, ,57 ± 0,12 28

29 outline workgroup instrumental analysis motivation photochemical synthesis of gold nanoparticle pre optimization DoE I optimization DoE II comparison experiment and simulation; outlook 29

30 experiments: summary and particularities procedure for experimental data perform first experiments to get an idea for the: input factors controllable non controllable: collect every possible measured values! output factors design a first design of experiments to: verify the factors and factor levels get a first idea of the system behavior detect regions in design space with failed designs second design of experiment with the aim: move design space to regions without failed designs to verify model of the system to increase dataset in the volume of interest 30

31 outlier detection outlier detection: important to maintain quality of optimization MOP- solver to test approximation quality (compare input (experimental) data with prediction of meta-model-of-optimal-prognosis) compare anthill plots or 3D approximation plots 3D- graph with approximation and marked outlier Anthill plot with marked outlier 31

32 outlook Transfer synthesis into an microfluidic set-up Improve prediction of nanoparticle size 32

33 Acknowledgements Instrumental Analysis Group Johannes Ott Bastian Böttcher Alexander Schmidt Maximilian Haschke Prof. Dr. Karl-Heinz Feller Dr. Jan-Steffen Niehaus Dr. Stephanie Kunath Michael Schimmelpfennig Funding 33

34 thank you! 34